The Silicon Valley Bank (SVB) was the 16th largest bank in the US. On 10 March 2023, it was put into receivership, merely 14 days after SVB’s auditor had issued a clean opinion on the financial statements closed on 31 December 2022. This has raised several questions:
In this article, we focus on the last element.
With the short-term funding collected from its depositors, that was repayable upon demand, SVB bought mainly long-term bonds, which it classified as “held to maturity” (HTM). In 2022, rising interest rates caused the bonds’ fair value to drop, but US GAAP rules say that SVB had not to recognize losses on these HTM bonds, as long as it did not sell them. With increasing withdrawals, SVB needed liquidity to pay them, which became increasingly difficult. Finally, on 10 March 2023, the regulator put SVB into receivership.
The extensive use of HTM accounting (HTM represented 43% of SVB’s total assets) allowed SVB to avoid recognizing unrealized losses in its earnings, equity and regulatory capital for several quarters, until massive withdrawals by its depositors made it inevitable that SVB would have to sell these securities to meet depositors’ demands.
SVB disclosed that the fair value of its HTM bonds was USD 76.2 billion at the end of 2022, or USD 15.1 billion below their book value. That unrealized loss was nearly as large as SVB’s USD 16.3 billion equity, as shown in the following extract of SVB’s 2022 financial statements.
Under current accounting rules in Topic 320, Investments-Debt Securities, HTM bonds are not required to be measured at fair value on the balance sheet, but should be reported at amortized cost. Bonds the bank may sell need to be classified as Available-for-Sale (AFS) and accounted for at fair value through Other Comprehensive Income (OCI). If a bank sells any HTM securities, all securities in that portfolio are considered “tainted” and have to be reclassified as AFS, leading the bank to potentially record a big loss on the securities it did not sell.
The use of the HTM accounting requires that the bank has not only the positive intent, but also the ability, to hold debt securities to maturity. The HTM classification is not permitted if a bank believes it may be required to sell a security for liquidity or other reasons.
As at 31 December 2022, SVB maintained its position that it intended and had the ability to hold those bonds to maturity. With the benefit of hindsight, it is clear that SVB might have had the intent but didn’t have the ability anymore to hold these bonds until their maturity. So, as SVB was going to be forced to sell part of its HTM bonds in 2023, this would have “tainted” its portfolio, and all the unrealized losses on the HTM bonds would have come to light…
This “mixed measurement approach”, where fair value accounting co-exists with the use of amortized cost, depending on the intent of the investor, is similar to the treatment under IAS 39, before IFRS 9 was implemented.
This discussion is not new. The issue came first under debate in the early 1990s, in the aftermath of the savings-and-loan crisis. Then, in 2010, in the fallout of the 2007-2009 global financial crisis, the Financial Accounting Standards Board (FASB) proposed requiring banks to record all financial instruments at fair value, to provide investors with more relevant information (much of the financial crisis resulted from illiquid, toxic subprime assets that had not been marked to market). But strong industry opposition caused the FASB to retreat in 2011 and stay with the mixed-measurement model.
The pros and cons of abandoning the hybrid model and putting all debt securities at fair value can be summarized as follows, based on the arguments of the various stakeholders:
Proponents of the full fair value approach claim that the fair value is the most relevant measure, as the management intent or the business model do not alter the value of a financial instrument.
According to them, fair values provide a better view of the performance of a bank’s investment strategy.
Some also argue that relying on the current management intent is questionable, as the management has a continuing obligation to reassess the most productive use of its assets, which is especially relevant considering the long-term nature of some bonds.
They also claim that valuing at fair value would improve consistency/comparability across reporting entities. And that the ability to choose a measurement basis provide opportunities for window dressing.
While the unrealized losses are not recognized, they are fully disclosed in the financial statements. Providing the fair value in the notes and expecting that users will reflect that in their analysis, puts less sophisticated investors at a disadvantage. This is why people sometimes refer to HTM as “Hide to Maturity”.
Others argue that management intent or business model matter, and should impact the reported value: when the objective of an investor is to collect the cash flows and it intends to keep the bond until its maturity, when it will recover (normally) the nominal value, they argue that amortized cost is the method which most closely reflects the economics of the strategy. Indeed, if a bond is kept to its maturity, the short-term differences between amortized cost and fair value eventually wash out…
Bankers generally consider that the Net Interest Margin is more relevant than the fair value in periods of market volatility.
Banks’ loan portfolios are not marked to market. Thus, it would not make sense for loans to not be marked to market, but to mark to market bonds presenting exactly the same risks.
Supporters of the mixed measurement approach also argue that measuring a large portion of the financial assets at fair value, while most liabilities would stay at amortized cost, would present a skewed picture of a bank’s balance sheet. And solving this by valuing the bank’s liabilities at fair value would be difficult, because the deposit franchise value is an intangible asset that is difficult to evaluate.
Following the global financial crisis of 2007-2009, some critics expressed the concern that the use of fair value accounting led some intermediaries to sell their positions, thereby further depressing the fair values, and creating a negative spiral, amplifying the crisis.
Finally, bankers argue that a wider use of the fair value measurement would increase the volatility of the net result and/or the equity, with irrelevant fair value fluctuations, when the bank intend to hold the debt securities until their maturity. This volatility would hurt a bank’s safety as investors might be discouraged from buying the bank’s stock (leading to higher cost of capital for banks), and depositors may be encouraged to withdraw their funds (increasing the liquidity risk to which a bank is exposed).
After the collapse of SVB, many European stakeholders raised the question: could the same happen to a European bank, considering the fact that European banks report under IFRS, and are subject to different regulatory requirements on liquidity risk and interest rate risk in the banking book.
Just like US GAAP, IFRS apply a mixed measurement model (although not identical), with some assets reported at amortized cost, and others at fair value. But, under IFRS, there is no “held to maturity” category anymore. Instead, IFRS 9 uses a “held to collect” business model, which is also measured at amortized cost, but subject to different conditions.
The business model assessment is performed based on scenarios that the bank reasonably expects to occur. This means the assessment excludes ‘stress case’ scenarios. For example, if a bank expects that it will sell a particular portfolio of bonds only in a stress case scenario, this would not affect the bank’s assessment of the business model for those bonds if it does not reasonably expect this to occur. And, if the bank sells more bonds than it expected when it classified them, this does not give rise to a prior period error in the bank’s financial statements, nor does it change the classification of the remaining bonds held in that business model. However, when the bank assesses the business model for newly purchased bonds, it must consider information about how cash flows were realised in the past, along with all other relevant information. For instance, if a business model changes from being held to collect due to increasing sales out of the portfolio, any new bonds recognised in that portfolio after such change would be classified as “held to collect and sell”, but existing bonds within the portfolio would continue to be measured at amortised cost.
This approach differs from the “tainting” rule applied under US GAAP. Also, it does not require that the bank assesses continuously whether it still has the ability to keep bonds until their maturity. In that sense, it could be argued that the IFRS rules are less severe than the US GAAP ones, and would not have triggered the accounting problem SVB experienced under US GAAP.
Also, it should be mentioned that, under IFRS 7, a Company must describe in the notes how it manages interest rate risk, and provide quantitative disclosures (a sensitivity analysis, or a duration mismatch, for instance). Such disclosures are however not very different from what is required under US GAAP.
But the main arguments making the SVB accident less likely in Europe are in the area of the supervisory tools put in place by the ECB: European banks are subjects to requirements in terms of liquidity risks and “interest rate risk in the banking book” which are more onerous than those which were imposed to SVB. This is the main reason why a SVB-like accident is less likely within a European bank.
The SVB collapse has revived the debate surrounding the “mixed measurement model”, allowing the coexistence of fair value accounting and amortized cost accounting for debt securities, under US GAAP.
Some commentators plead for a radical solution: eliminating the HTM category and putting all debt securities (or even all financial assets or all financial instruments) at fair value in the financial statements.
But there is a more easy solution. We should remind that the conditions for using the HTM category is that the company has not only the positive intent, but also the ability, to hold debt securities to their maturity. In the case at hand, it is clear that SVB didn’t have (anymore) that ability, and did not draw the conclusion thereof: the need to reclassify a large part of its HTM portfolio to the FV category. So, a rigorous application by SVB of the existing rules could have put SVB’s problem in light much earlier.
The accounting regulator could also provide guidance on the criteria that should substantiate a bank’s assertion that it is able to hold debt securities to maturity, and could even force firms to provide in their financial statements evidence that they are actually able to hold their HTM bonds to maturity.
“At the end of 2022, the unrealized loss on SVB’s HTM bonds was nearly as large as its equity.”
“As at 31 December 2022, SVB might have had the intent but had not anymore the ability to hold these bonds until their maturity.”
“A rigorous application by SVB of the existing rules could have put SVB’s problem in light much earlier.”
Joao Granja, “Bank fragility and reclassification of securities into HTM”, University of Chicago, 2023
Mark Mauer, “Banks, investors revive push for changes to securities accounting after SVB collapse”, Wall Street Journal, 20 March 2023
Michael J. Walker, “Accounting for debt securities in the age of Silicon Valley Bank”, Federal Reserve Bank of Boston, SRA Notes, 5 October 2023
Michael S. Barr, “Review of the Federal Reserve’s Supervision and Regulation of Silicon Valley Bank”, Board of Governors of the Federal Reserve System, 28 April 2023
Sandy Peters, “The SVB collapse: FASB should eliminate “Hide-‘til-Maturity” accounting”, CFA Institute, 2023
S.J. Steinhardt, “Bank collapse prompts another look at fair-value accounting for securities”, New York State Society of CPAs, 2023
The Footnotes Analyst, “Fair values and interest rate risk – Silicon Valley Bank”